For many years, it was of interest to identify the sequences encoding the two melatonin receptors (MT₁ and MT₂) from various species. After publishing the basic molecular characterization of the human, rat, mouse, sheep, and platypus MT₁, MT₂, or Mel1c receptors, we began cloning the genes from other animals, such as birds, bats, and vipers. The goal was to advance the receptor crystallization, which could greatly contribute the understanding of the sequence/stability relationship. European hamster MT₁ receptor was cloned for the first time from this gender, was expressed in stable form in cells, and its binding characterized with a sample of 19 melatonin ligands. Siberian hamster (Phodopus sungorus) expresses a non-functional MT₂. We observed that unlike this hamster, the European hamster (Cricetus cricetus) does not have a stop codon in the MT₂ sequence. Thus, we undertook the tedious task of cloning the MT₂ receptor. We partially succeeded, sequencing the complete exon 2 and a fragment of exon 1 (from putative amino acids 12 to 38 and 77 to 323), after several years of efforts. In order to show that the protein parts we cloned were capable to sustain some binding capacities, we designed a chimeric MT₂ receptor using a consensus sequence to replace the unknown amino acids, based on other small rodent MT₂ sequences. This chimeric construct could bind melatonin in the nanomolar range. This work is meant to be the basis for attempts from other laboratories of the community to determine the complete natural sequence of the European hamster MT₂ receptor. The present work is the first to show that, among the hamsters, if the Siberian is a natural knockout for MT₂, the European one is not.

Predicting fluorescently labelled cellular structures from brightfield images is a recent application of convolutional neural networks and has already proven a valuable tool in biological imaging studies. These methods reduce the need for time-consuming manual annotations for supervised datasets, potentially reduce the costs in high-throughput imaging screens, and open up a number of potentially novel analyses. However as with any prediction there can be sources of error and uncertainty. Here we present BFNet, a method to visualise the uncertainty in predicted images by applying Monte Carlo dropout during inference and calculating per-pixel variance as an uncertainty map. Our method demonstrates the ability to highlight regions of an image where prediction is difficult or impossible due to imaging artefacts such as occlusions or out-of-focus images, as well as more general uncertainty when a trained model is applied to new data from different imaging of experimental settings. We have provided a python implementation of the method which is available at github.com/swarchal/bfnet .